In the era of evidence-based medicine we often fall into the trap of evaluating the performance of an individual provider or a healthcare system by the degree to which they comply with clinical guidelines or best practices that are grounded in outcomes from randomized controlled trials. Yet, when we are asked by a friend or family member about a referral to an individual physician or a hospital, we always have a few that we preferentially recommend – even though there are others that perform equally well (or even better) on the measures mentioned above. In effect, we are taking into account the “art” as well as the “science” of medicine.
Enter a very humbling paper (Weiner et al 2010) on “contextual” errors. Contextual errors are those errors where the clinician “fails to take into account those elements of a patient’s environment or behavior that are relevant to their care”. Examples of those elements might be a patient’s financial or employment status, social situation, health literacy, etc. How many times have you “escalated” a patient’s antihypertensive regimen, only to realize later that they were never compliant with any part of that regimen, either because they could not afford those medications or could not read the labels? That’s an example of a contextual error.
The study by Weiner et al. puts the frequency of such contextual errors in perspective and demonstrates why we so often fail to achieve the desired outcome despite “adhering to the guidelines”. It’s also a good reminder that the idealized “subjects” in randomized controlled trials have been carefully selected to weed out many of these contextual issues and may explain some of the differences between clinical trial and “real world” outcomes.
The investigators trained actors in variants of 4 common clinical scenarios and sent them as unannounced standardized patients into the practices of 111 internal medicine attending physicians. For each scenario there were “red flags” that required further probing by the clinician. Red flags were either “contextual” (as above) or “biomedical” (for example, a nocturnal increase in wheezing or coughing that should prompt questions about possible esophageal reflux). So each scenario could have no red flags, a contextual red flag, a biomedical red flag, or both red flags.
As we’d suspect, fewer clinicians probed further after the “contextual” red flags, but the impact on overall plans of care was striking. Error-free plans of care occurred in 73% of the “uncomplicated” (no red flags) cases but in only 38% of the biomedically complicated cases and 22% of the contextually complicated cases. And of those with both biomedical and contextual red flags a mere 9% of cases had error-free plans of care.
What a powerful demonstration that, as we strive to standardize care, we must not sacrifice the need to individualize care for each patient.
While we can program electronic medical records to remind you to add an ACE inhibitor is certain clinical situations, it is much more difficult to rely on technological solutions to address contextual issues. We need to do a better job of training our medical students and residents to consider these contextual issues as they interact with patients and their families. Most medical schools now include simulations (using actors or standardized patients) in teaching interview techniques. It would be easy to add the sort of contextual issues from the Weiner paper to those simulation training exercises.
But we have some other suggestions as well. The successful clinician is one who utilizes all members of the healthcare team in dealing with his or her patients. It is amazing how often a patient will confide some of these contextual issues to a nurse or clerical staff but would not disclose them to a physician. Sometimes they are too embarrassed to discuss them with the physician. Other times they may feel that the physician is “too busy” and they don’t want to bother them with these issues. So sometimes it is okay to have one of your other team members broach the questions for you. (As an aside, we’ve seen many teens who are afraid to talk about issues like STD’s but will listen to tapes on such topics or even talk to anonymous sources about such issues. Generation X also freely uses social media technology to discuss things they would never discuss in person.). The solution is that you need to have multiple means of communication available to your patients because they may all communicate in different ways.
Contextual issues can also pop up when you are doing team meetings in your practices. While most practices focus on issues such as billing, scheduling, etc. during “team meetings”, that’s also a good time to say “We’re having trouble getting Mrs. Jones’ diabetes under control. Does anyone have any insights that might help us better manage her care?”. You’d be surprised how often your staff will volunteer their insights and it also helps your staff take pride in helping all your patients.
You also need to include contextual issues in your plans of care. For example, even if you neglected to probe contextual issues when you first developed a plan of care, you should always ask the appropriate contextual questions when a patient has not responded as expected to a therapeutic intervention. While you should have a clear understanding about the financial impact of prescribing any new drug for a patient, it becomes absolutely crucial that you specifically inquire about cost issues if their blood pressure has not improved or their LDL has not moved a month after you started them on a new regimen.
It can be a humbling experience when you achieve a less than desired outcome because you didn’t take the time to give your patient the individualized care he/she needed. This study by Weiner et al. is a real reminder that we still take care of patients one at a time and it’s a real contribution to helping us deliver safe and effective care.
Saul J. Weiner SJ, Schwartz A, Weaver F, et al. Contextual Errors and Failures in Individualizing Patient Care: A Multicenter Study. Ann Intern Med 2010; 153: 69-75